Mailing List Archive

Hey, anyone who wanna help improve the net stability?
Hey all!

We are approaching 40K routes, and this seems to make IBGP in networks
with rich external connectivity pretty unhappy.

Without pointing fingers, there are some 9K prefixes that could be
removed by proper aggregation/reconfiguration.

ICM-Stockholm-2#sh ip bgp summ
BGP table version is 526856, main routing table version 526856
39321 network entries (300167/309954 paths) using 13491704 bytes of memory


I suggest that everyon go look at what they can do, as we are all
loosers with the currect situation.

--Peter
- - - - - - - - - - - - - - - - -
Re: Hey, anyone who wanna help improve the net stability? [ In reply to ]
Peter,
Is there a way we can help?
We are doing what we can to aggregate.
Bob

On Thu, 5 Sep 1996, Peter Lothberg wrote:

>
>
> Hey all!
>
> We are approaching 40K routes, and this seems to make IBGP in networks
> with rich external connectivity pretty unhappy.
>
> Without pointing fingers, there are some 9K prefixes that could be
> removed by proper aggregation/reconfiguration.
>
> ICM-Stockholm-2#sh ip bgp summ
> BGP table version is 526856, main routing table version 526856
> 39321 network entries (300167/309954 paths) using 13491704 bytes of memory
>
>
> I suggest that everyon go look at what they can do, as we are all
> loosers with the currect situation.
>
> --Peter
>

- - - - - - - - - - - - - - - - -
Re: Hey, anyone who wanna help improve the net stability? [ In reply to ]
> Without pointing fingers, there are some 9K prefixes that could be
> removed by proper aggregation/reconfiguration.

There are also 25,000 /24's, up 20% from six months ago. I don't
have any figures on how many are in the "swamp", but I think this
is the most fruitful place to try to cap growth or hack it back:

Total routes: 39670

8 22 0.1%
9 1 0.0%
10 4 0.0%
11 6 0.0%
12 14 0.0%
13 30 0.1%
14 102 0.3%
15 173 0.4%
16 5687 14.3%
17 246 0.6%
18 493 1.2%
19 921 2.3%
20 954 2.4%
21 1200 3.0%
22 1824 4.6%
23 2425 6.1%
24 25539 64.4%
...
- - - - - - - - - - - - - - - - -
Re: Hey, anyone who wanna help improve the net stability? [ In reply to ]
Hi Paul,

you comment leaves me confused. 25,000 24s up 20% in the last six months
means 5,000 NEW prefix 24s in the global routing tables. Where did they
come from? I ask because I had thought that it was just about impossible
to get a 24 routed at the nets defaultless core. And that this
impossibility has been around for at least the last 6 months.

I thought that if "cooknet" as a new MCI customer has a 24 handed to it by
mci all nice and cidrized that "cooknet's" 24 would never appear in your
list being aggregated by mci along with other 24s to make a smaller prefix
that would be announced eventually in the global tables? Have I
misunderstood something or is theory diverging from practice.

************************************************************************
The COOK Report on Internet For subsc. pricing & more than
431 Greenway Ave, Ewing, NJ 08618 USA ten megabytes of free material
(609) 882-2572 (phone & fax) visit http://pobox.com/cook/
Internet: cook@cookreport.com For case study of MercerNet &
TIIAP induced harm to local community http://pobox.com/cook/mercernet.html
************************************************************************


On Thu, 5 Sep 1996, Paul A Vixie wrote:

> > Without pointing fingers, there are some 9K prefixes that could be
> > removed by proper aggregation/reconfiguration.
>
> There are also 25,000 /24's, up 20% from six months ago. I don't
> have any figures on how many are in the "swamp", but I think this
> is the most fruitful place to try to cap growth or hack it back:
>
> Total routes: 39670
>
> 8 22 0.1%
> 9 1 0.0%
> 10 4 0.0%
> 11 6 0.0%
> 12 14 0.0%
> 13 30 0.1%
> 14 102 0.3%
> 15 173 0.4%
> 16 5687 14.3%
> 17 246 0.6%
> 18 493 1.2%
> 19 921 2.3%
> 20 954 2.4%
> 21 1200 3.0%
> 22 1824 4.6%
> 23 2425 6.1%
> 24 25539 64.4%
> ...
>

- - - - - - - - - - - - - - - - -
Re: Hey, anyone who wanna help improve the net stability? [ In reply to ]
This is timely for me. I recently picked up a customer who became duel
homed by virtue of their circuit to me. This customer had been propagating
145 routes via their other provider. With aggregation they only needed to
propagate 34. After having the situation explained, the customer was open
to fixing their config. Providers should, better than anyone else, understand
the benefits of minimizing the size of the global routing table. We could
substantially reduce its size by just educating/policing our customers.

Jim
>
>
>
> Hey all!
>
> We are approaching 40K routes, and this seems to make IBGP in networks
> with rich external connectivity pretty unhappy.
>
> Without pointing fingers, there are some 9K prefixes that could be
> removed by proper aggregation/reconfiguration.
>
> ICM-Stockholm-2#sh ip bgp summ
> BGP table version is 526856, main routing table version 526856
> 39321 network entries (300167/309954 paths) using 13491704 bytes of memory
>
>
> I suggest that everyon go look at what they can do, as we are all
> loosers with the currect situation.
>
> --Peter
>

- - - - - - - - - - - - - - - - -
Re: Hey, anyone who wanna help improve the net stability? [ In reply to ]
Umm, Gordon, read what Paul said.

He was saying that he suspected that the new /24s are in the swamp.
And even Sprint listens to any /24s from there.

The swamp is basically the old "Classful" address space.
What Sean did was to say "You can use your old 'Class C's
but I have to stop the table-size growth in *new allocations*".

Avi

> Hi Paul,
>
> you comment leaves me confused. 25,000 24s up 20% in the last six months
> means 5,000 NEW prefix 24s in the global routing tables. Where did they
> come from? I ask because I had thought that it was just about impossible

They either came from:

1) The swamp, or
2) Sprintlink customers

Sean's answer to 2) would be that other providers should adopt similar
filters.

> to get a 24 routed at the nets defaultless core. And that this
> impossibility has been around for at least the last 6 months.
>
> I thought that if "cooknet" as a new MCI customer has a 24 handed to it by
> mci all nice and cidrized that "cooknet's" 24 would never appear in your
> list being aggregated by mci along with other 24s to make a smaller prefix

True if MCI aggregated properly. But if you're dual-homed to Sprintlink,
Sprint'd announce it for you - and if you used a 2-year-old "Class C"
obtained for "cooknet" from the NIC, Sprint would hear that and/or
announce it for you (depending on whether you're a customer, of course).

> that would be announced eventually in the global tables? Have I
> misunderstood something or is theory diverging from practice.

Avi

> ************************************************************************
> The COOK Report on Internet For subsc. pricing & more than
> 431 Greenway Ave, Ewing, NJ 08618 USA ten megabytes of free material
> (609) 882-2572 (phone & fax) visit http://pobox.com/cook/
> Internet: cook@cookreport.com For case study of MercerNet &
> TIIAP induced harm to local community http://pobox.com/cook/mercernet.html
> ************************************************************************
>
>
> On Thu, 5 Sep 1996, Paul A Vixie wrote:
>
> > > Without pointing fingers, there are some 9K prefixes that could be
> > > removed by proper aggregation/reconfiguration.
> >
> > There are also 25,000 /24's, up 20% from six months ago. I don't
> > have any figures on how many are in the "swamp", but I think this
> > is the most fruitful place to try to cap growth or hack it back:
> >
> > Total routes: 39670
> >
> > 8 22 0.1%
> > 9 1 0.0%
> > 10 4 0.0%
> > 11 6 0.0%
> > 12 14 0.0%
> > 13 30 0.1%
> > 14 102 0.3%
> > 15 173 0.4%
> > 16 5687 14.3%
> > 17 246 0.6%
> > 18 493 1.2%
> > 19 921 2.3%
> > 20 954 2.4%
> > 21 1200 3.0%
> > 22 1824 4.6%
> > 23 2425 6.1%
> > 24 25539 64.4%
> > ...
> >
>

- - - - - - - - - - - - - - - - -
Re: Hey, anyone who wanna help improve the net stability? [ In reply to ]
Avi pointed out:

The swamp is basically the old "Classful" address space.
What Sean did was to say "You can use your old 'Class C's
but I have to stop the table-size growth in *new allocations*".

=========
Then apparently I have misunderstood what the swamp was/is. I thought it
was just an unaggregated area and didn't understand why. Is there text
somewhere that explains very clearly the swamp and the policies attached
there to?

Is the swamp then bounded by class c addresses warranted as routable when
they were handed out? Are you saying then that the defaultless core
routability of class c's from the swamp is, as of now, guaranteed?

Of the class cs from the swamp how many are now being routed at the
defaultless core? HOW MANY ADDITIONAL CLASS Cs FROM THE SWAMP ARE THERE
FOR WHICH THE OWNERS COULD DEMAND ROUTING?

In other words are these 5,000 new class c's just the beginning? Or are
they, hopefully the end?
=========

Avi writes:

Umm, Gordon, read what Paul said.

He was saying that he suspected that the new /24s are in the swamp.
And even Sprint listens to any /24s from there.

The swamp is basically the old "Classful" address space.
What Sean did was to say "You can use your old 'Class C's
but I have to stop the table-size growth in *new allocations*".

Avi


************************************************************************
The COOK Report on Internet For subsc. pricing & more than
431 Greenway Ave, Ewing, NJ 08618 USA ten megabytes of free material
(609) 882-2572 (phone & fax) visit http://pobox.com/cook/
Internet: cook@cookreport.com For case study of MercerNet &
TIIAP induced harm to local community http://pobox.com/cook/mercernet.html
************************************************************************


- - - - - - - - - - - - - - - - -
Re: Hey, anyone who wanna help improve the net stability? [ In reply to ]
> Avi pointed out:
>
> The swamp is basically the old "Classful" address space.
> What Sean did was to say "You can use your old 'Class C's
> but I have to stop the table-size growth in *new allocations*".
>
> =========
> Then apparently I have misunderstood what the swamp was/is. I thought it
> was just an unaggregated area and didn't understand why. Is there text
> somewhere that explains very clearly the swamp and the policies attached
> there to?

It's not necessarily unaggregated...

I'm not sure, it's possible that Sean might have described the swamp as
basically space above the a/b space and < 205/8.

> Is the swamp then bounded by class c addresses warranted as routable when
> they were handed out? Are you saying then that the defaultless core

No, warranted as "ok, I can deal with routes from that space" by Sprint's
filters.

> routability of class c's from the swamp is, as of now, guaranteed?

Define "guarantee". No, nothing is "guaranteed", but there'd be HUGE
objection to any new filtering policies that affected already-allocated-
and-routed space.

> Of the class cs from the swamp how many are now being routed at the
> defaultless core? HOW MANY ADDITIONAL CLASS Cs FROM THE SWAMP ARE THERE

No clue, but we could find out.

> FOR WHICH THE OWNERS COULD DEMAND ROUTING?

Many :)

> In other words are these 5,000 new class c's just the beginning? Or are
> they, hopefully the end?

Hard to say, we'd have to see how many of the new /24s are actually
"Class C"s.

Avi

- - - - - - - - - - - - - - - - -
Re: Hey, anyone who wanna help improve the net stability? [ In reply to ]
On Thu, 5 Sep 1996, Gordon Cook wrote:

==>Is the swamp then bounded by class c addresses warranted as routable when
==>they were handed out? Are you saying then that the defaultless core
==>routability of class c's from the swamp is, as of now, guaranteed?

They were never warranted or guaranteed as routable. However, the NIC
didn't place any warnings that space smaller than a certain amount may not
be routable, and at the time, no one had placed filters on route
distribution like SprintLink's. The NIC opted to revise its wording on
the application around the same time SprintLink placed its filter on new
allocations.

For right now, the routability of class C's is not guaranteed out of the
swamp; however, I know of no route filters out there which will restrict
them.

/cah


- - - - - - - - - - - - - - - - -
Re: Hey, anyone who wanna help improve the net stability? [ In reply to ]
Gordon, please make your messages easier to read or people will /dev/null
them. It's really not clear what, in the message below, are the new
questions/points...

Avi

> Avi pointed out:
>
> The swamp is basically the old "Classful" address space.
> What Sean did was to say "You can use your old 'Class C's
> but I have to stop the table-size growth in *new allocations*".
>
> =========
> Then apparently I have misunderstood what the swamp was/is. I thought it
> was just an unaggregated area and didn't understand why. Is there text
> somewhere that explains very clearly the swamp and the policies attached
> there to?
>
> Is the swamp then bounded by class c addresses warranted as routable when
> they were handed out? Are you saying then that the defaultless core
> routability of class c's from the swamp is, as of now, guaranteed?
>
> Of the class cs from the swamp how many are now being routed at the
> defaultless core? HOW MANY ADDITIONAL CLASS Cs FROM THE SWAMP ARE THERE
> FOR WHICH THE OWNERS COULD DEMAND ROUTING?
>
> In other words are these 5,000 new class c's just the beginning? Or are
> they, hopefully the end?
> =========
>
> Avi writes:
>
> Umm, Gordon, read what Paul said.
>
> He was saying that he suspected that the new /24s are in the swamp.
> And even Sprint listens to any /24s from there.
>
> The swamp is basically the old "Classful" address space.
> What Sean did was to say "You can use your old 'Class C's
> but I have to stop the table-size growth in *new allocations*".
>
> Avi
>
>
> ************************************************************************
> The COOK Report on Internet For subsc. pricing & more than
> 431 Greenway Ave, Ewing, NJ 08618 USA ten megabytes of free material
> (609) 882-2572 (phone & fax) visit http://pobox.com/cook/
> Internet: cook@cookreport.com For case study of MercerNet &
> TIIAP induced harm to local community http://pobox.com/cook/mercernet.html
> ************************************************************************
>
>

- - - - - - - - - - - - - - - - -
Re: Hey, anyone who wanna help improve the net stability? [ In reply to ]
Original message <Pine.SUN.3.94.960905145324.10907T-100000@unix1.netaxs.com>
From: Gordon Cook <cook@netaxs.com>
Date: Sep 5, 15:04
Subject: Re: Hey, anyone who wanna help improve the net stability?
...
>
> Of the class cs from the swamp how many are now being routed at the
> defaultless core? HOW MANY ADDITIONAL CLASS Cs FROM THE SWAMP ARE THERE
> FOR WHICH THE OWNERS COULD DEMAND ROUTING?
...

Exactly as many class C's as are not yet being routed, minus those which
have been voluntarily returned. I believe the return statistics have been
made available to nanog before, so perhaps the keeper of them could do
this calculation.

-matthew kaufman
matthew@scruz.net

- - - - - - - - - - - - - - - - -
Re: Hey, anyone who wanna help improve the net stability? [ In reply to ]
Avi,

> It's not necessarily unaggregated...
>
> I'm not sure, it's possible that Sean might have described the swamp as
> basically space above the a/b space and < 205/8.

At the Los Angeles IETF I tried to define the terms "Swamp" and "Toxic
Waste Dump" as follows:
* The Swamp is the 192.0.0.0/8 addres space that is routed.
* Each and every /24 in the Swamp is part of the Toxic Waste Dump (twd).

> > Of the class cs from the swamp how many are now being routed at the
> > defaultless core? HOW MANY ADDITIONAL CLASS Cs FROM THE SWAMP ARE THERE
>
> No clue, but we could find out.

You might not like/agree/understand the above definitions, but... given
the above definitions the current sizes are:

Size(swamp) = 6549
Size(twd) = 5555

This data is taken from a default free router that lives in Amsterdam,
The Netherlands which has 40590 prefixes at this moment.

FYI.

__

Erik-Jan.
- - - - - - - - - - - - - - - - -
Re: Hey, anyone who wanna help improve the net stability? [ In reply to ]
Below...

......... Gordon Cook is rumored to have said:
]
] Avi pointed out:
]
] The swamp is basically the old "Classful" address space.
] What Sean did was to say "You can use your old 'Class C's
] but I have to stop the table-size growth in *new allocations*".
]
] =========
] Then apparently I have misunderstood what the swamp was/is. I thought it
] was just an unaggregated area and didn't understand why. Is there text
] somewhere that explains very clearly the swamp and the policies attached
] there to?

I see a question that says why was it not aggregated?

It wasn't aggregate because at the time of dispersion, there was
no need to allocate, routing table growth was not a significant
factor in internet address allocation.

] Is the swamp then bounded by class c addresses warranted as routable when
] they were handed out? Are you saying then that the defaultless core
] routability of class c's from the swamp is, as of now, guaranteed?

Perhaps I'm a bit of a wild card, but I don't see any IP address
as routable based on official policy or origin of assignment. IP
Address space is routable because each [I|N]SP that might have
occasion to route the space, agrees to. They agree to because it
makes economic sense, or, in some cases *cough sprint* it does not
make economic sense to route 207.99.99.0/24 (apologies to NAC.NET,
I just made the net up, and I know they have a /18, so the example
is invalid...) . I firmly believe that explainable economic decisions
are in the best interest of the net, and if they're not, the
company making poor economic decisions will leave through natural
[economic] selection.

] Of the class cs from the swamp how many are now being routed at the
] defaultless core? HOW MANY ADDITIONAL CLASS Cs FROM THE SWAMP ARE THERE
] FOR WHICH THE OWNERS COULD DEMAND ROUTING?

Lots.

] In other words are these 5,000 new class c's just the beginning? Or are
] they, hopefully the end?

Neither, they're a holdover from a more carefree time...

At least, that's my 3 minute rant before my next meeting... :)

-alan

] =========
]
] Avi writes:
]
] Umm, Gordon, read what Paul said.
]
] He was saying that he suspected that the new /24s are in the swamp.
] And even Sprint listens to any /24s from there.
]
] The swamp is basically the old "Classful" address space.
] What Sean did was to say "You can use your old 'Class C's
] but I have to stop the table-size growth in *new allocations*".
]
] Avi
]
]
] ************************************************************************
] The COOK Report on Internet For subsc. pricing & more than
] 431 Greenway Ave, Ewing, NJ 08618 USA ten megabytes of free material
] (609) 882-2572 (phone & fax) visit http://pobox.com/cook/
] Internet: cook@cookreport.com For case study of MercerNet &
] TIIAP induced harm to local community http://pobox.com/cook/mercernet.html
] ************************************************************************
]
]
]

- - - - - - - - - - - - - - - - -
Re: Hey, anyone who wanna help improve the net stability? [ In reply to ]
At 03:04 PM 9/5/96 -0400, Gordon Cook wrote:

>Then apparently I have misunderstood what the swamp was/is. I thought it
>was just an unaggregated area and didn't understand why. Is there text
>somewhere that explains very clearly the swamp and the policies attached
>there to?
>

Gordon,

The Swamp is 192/8. The unnaggregated /24 allocations in The Swamp is
called The Toxic Waste Dump [TWD].

Also see:

http://www.isi.edu/div7/pier/whose_routes

- paul

- - - - - - - - - - - - - - - - -
Re: Hey, anyone who wanna help improve the net stability? [ In reply to ]
> You might not like/agree/understand the above definitions, but... given
> the above definitions the current sizes are:

I understand the above definitions, of course...

> Size(swamp) = 6549
> Size(twd) = 5555
>
> This data is taken from a default free router that lives in Amsterdam,
> The Netherlands which has 40590 prefixes at this moment.
>
> FYI.
>
> Erik-Jan.

Avi

- - - - - - - - - - - - - - - - -
Re: Hey, anyone who wanna help improve the net stability? [ In reply to ]
Erik-Jan

You were plotting the size of the routing table at very tight
intervals (1 hour or so as I recall) - are these graphs online
somewhere?

geoff

>
> Avi,
>
> > It's not necessarily unaggregated...
> >
> > I'm not sure, it's possible that Sean might have described the swamp as
> > basically space above the a/b space and < 205/8.
>
> At the Los Angeles IETF I tried to define the terms "Swamp" and "Toxic
> Waste Dump" as follows:
> * The Swamp is the 192.0.0.0/8 addres space that is routed.
> * Each and every /24 in the Swamp is part of the Toxic Waste Dump (twd).
>
> > > Of the class cs from the swamp how many are now being routed at the
> > > defaultless core? HOW MANY ADDITIONAL CLASS Cs FROM THE SWAMP ARE THERE
> >
> > No clue, but we could find out.
>
> You might not like/agree/understand the above definitions, but... given
> the above definitions the current sizes are:
>
> Size(swamp) = 6549
> Size(twd) = 5555
>
> This data is taken from a default free router that lives in Amsterdam,
> The Netherlands which has 40590 prefixes at this moment.
>
> FYI.
>
> __
>
> Erik-Jan.
>
>

- - - - - - - - - - - - - - - - -
Re: Hey, anyone who wanna help improve the net stability? [ In reply to ]
Now that net99 is assimilated (with the exception of one city), AGIS will
begin cleaning up its announcements, as well as implementing access-list 112
inbound. We'll publish the dates and details soon.

Peter



At 02:51 PM 9/5/96 DST, you wrote:
>
>
>Hey all!
>
>We are approaching 40K routes, and this seems to make IBGP in networks
>with rich external connectivity pretty unhappy.
>
>Without pointing fingers, there are some 9K prefixes that could be
>removed by proper aggregation/reconfiguration.
>
>ICM-Stockholm-2#sh ip bgp summ
>BGP table version is 526856, main routing table version 526856
>39321 network entries (300167/309954 paths) using 13491704 bytes of memory
>
>
>I suggest that everyon go look at what they can do, as we are all
>loosers with the currect situation.
>
>--Peter
>
>

_____________________________________________________________________
Peter Kline | 313-730-5151
AGIS - Internet Backbone Services | _Lucem Diffundo_
Post-Traumatic Success Disorder+ |
/////////////////////////////////////////////////////////////////////
You can pretend to care, but you can't pretend to be there.

- - - - - - - - - - - - - - - - -
Re: Hey, anyone who wanna help improve the net stability? [ In reply to ]
Geoff,

> You were plotting the size of the routing table at very tight
> intervals (1 hour or so as I recall) - are these graphs online
> somewhere?

Every hour on the hour I take a sample.

The latest and greatest graph can be found at (sorry, no gif or jpeg
available currently):
ftp://ftp.nic.surfnet.nl/surfnet/net-management/ip/nets.ps

Please note the flat-line at the end of the graph: Due to a reconfig of
the router I take the samples from during my holidays I am missing some
data points :-(. But the last sample points that are in there are well
above 40000.

__

Erik-Jan.
- - - - - - - - - - - - - - - - -
Re: Hey, anyone who wanna help improve the net stability? [ In reply to ]
> There are also 25,000 /24's, up 20% from six months ago. I don't
> have any figures on how many are in the "swamp", but I think this
> is the most fruitful place to try to cap growth or hack it back:

I agree. Ideally there can be some type of gentle persuasion to make this
happen - ie, if you renumber from a /24, you will get better
reliability. Sprint's dampening policy encourages this.

> 23 2425 6.1%
> 24 25539 64.4%
> ...

- - - - - - - - - - - - - - - - -
Re: Hey, anyone who wanna help improve the net stability? [ In reply to ]
> Of the class cs from the swamp how many are now being routed at the
> defaultless core? HOW MANY ADDITIONAL CLASS Cs FROM THE SWAMP ARE THERE
> FOR WHICH THE OWNERS COULD DEMAND ROUTING?
>
> In other words are these 5,000 new class c's just the beginning? Or are
> they, hopefully the end?

Lets see, what the BOE math tell us...

24,000 entries now,
192.0.0.0/8 is ~2million.

~2mil - 24,000 = current router death.

--
--bill
- - - - - - - - - - - - - - - - -
RE: Hey, anyone who wanna help improve the net stability? [ In reply to ]
On Thursday, September 05, 1996 9:52 AM, bmanning@isi.edu wrote:
@ > Of the class cs from the swamp how many are now being routed at the
@ > defaultless core? HOW MANY ADDITIONAL CLASS Cs FROM THE SWAMP ARE THERE
@ > FOR WHICH THE OWNERS COULD DEMAND ROUTING?
@ >
@ > In other words are these 5,000 new class c's just the beginning? Or are
@ > they, hopefully the end?
@
@ Lets see, what the BOE math tell us...
@
@ 24,000 entries now,
@ 192.0.0.0/8 is ~2million.
@
@ ~2mil - 24,000 = current router death.
@
@ --
@ --bill
@
@

Is now when we say..."we told you so..."....???

You are seeing the great allocation policies and procedures at work...

fortunately, there will be other address spaces and the IPv4 space
can be wrapped up as one TWD...

--
Jim Fleming
UNETY Systems, Inc.
Naperville, IL

e-mail: JimFleming@unety.net

- - - - - - - - - - - - - - - - -
RE: Hey, anyone who wanna help improve the net stability? [ In reply to ]
[cc: list mercifully trimmed]

Jim,

Regadless of whether the address space is IPv4 or IPv6 (or IPv8 or
whatever nonsense), the same problem would exist with this number
of prefixes. This is *not* an issue of 'which address space is
hosed', but rather, how to put into place practical allocation and
aggregation strategies to keep the total number of prefixes to
a minimum.

It would appear that while you seem overly anxious to say 'I told
you so', you've completely ignored the technical discussions on
why the problem exists in the first place, and why the allocation
and aggregation methodologies need vast improvement than history
has shown us.

This is completely orthogonal to 'other address spaces'.

- paul

At 05:15 PM 9/5/96 -0500, Jim Fleming wrote:

>
>Is now when we say..."we told you so..."....???
>
>You are seeing the great allocation policies and procedures at work...
>
>fortunately, there will be other address spaces and the IPv4 space
>can be wrapped up as one TWD...
>

- - - - - - - - - - - - - - - - -
Re: Hey, anyone who wanna help improve the net stability? [ In reply to ]
> On Thursday, September 05, 1996 9:52 AM, bmanning@isi.edu wrote:
> @ > Of the class cs from the swamp how many are now being routed at the
> @ > defaultless core? HOW MANY ADDITIONAL CLASS Cs FROM THE SWAMP ARE THERE
> @ > FOR WHICH THE OWNERS COULD DEMAND ROUTING?
> @ >
> @ > In other words are these 5,000 new class c's just the beginning? Or are
> @ > they, hopefully the end?
> @
> @ Lets see, what the BOE math tell us...
> @
> @ 24,000 entries now,
> @ 192.0.0.0/8 is ~2million.
> @
> @ ~2mil - 24,000 = current router death.
> @
> @ --
> @ --bill

Wait, 192/8 contains 2^(16-8=8) = 256 /16s.
Each /16 contains 2^(24-16=8) = 256 /24s.
256 * 256 = 65536 (give or take a few which "mean something" special).
65536-24000=41536.

Not pretty, but not definitively router death.

> Is now when we say..."we told you so..."....???

Who told who what?

> You are seeing the great allocation policies and procedures at work...

So are you of the "one-route-per-provider, but only 10-20 provider"
school of thought?

> fortunately, there will be other address spaces and the IPv4 space
> can be wrapped up as one TWD...
>
> --
> Jim Fleming

Avi
- - - - - - - - - - - - - - - - -
Re: Hey, anyone who wanna help improve the net stability? [ In reply to ]
At 03:01 PM 9/5/96 -0500, Alan Hannan wrote:

>] Of the class cs from the swamp how many are now being routed at the
>] defaultless core? HOW MANY ADDITIONAL CLASS Cs FROM THE SWAMP ARE THERE
>] FOR WHICH THE OWNERS COULD DEMAND ROUTING?
>
> Lots.

Uhm, I dunno that people can really /demand/ routing. Ask for it, expect it
or pray for it yes, but there is no reason that ISP X won't come up with a
policy tomorrow saying we have a policy saying that we won't route any
addresses that come from Justin.

Justin Newton
Internet Architect
Erol's Internet Services

- - - - - - - - - - - - - - - - -
RE: Hey, anyone who wanna help improve the net stability? [ In reply to ]
It seems the problem is number of flaps, etc. Doesn't it make sense to
simply start dropping the routes that flap the most? Of course this will
be biased in the /24 space but at least the rationale/solution maps
identically to a solid reason.

To what extent is this flapping due to basically static information being
propogated? One solution would be to nail these routes ...

Cheers, peter


----------
- - - - - - - - - - - - - - - - -

1 2  View All